Method to detect and score churn in online social games

Abstract

A method and a system for predicting churn of a player of an online game is described. Online engagements of a group of players of the online game are monitored during a churn prediction model training period. Online engagement scores for the group of players are computed within a periodic number of days within the churn model training period. A weighted exponential moving average of the online engagement scores of the group of players of the online game is computed during the churn model training period. The weighted exponential moving average is used to determine an online engagement threshold value of a churn prediction model for the online game. The online engagement threshold value is applied to a weighted exponential moving average of a player during an observation period to determine a churn probability of the player within a prediction period.

What is claimed is: 
     
       1. A system comprising:
 an online engagement monitoring module comprising one or more computer processors configured to monitor online engagements of a plurality of players of a first online game, to determine respective engagement parameters for each player; and 
 a recommendation module comprising one or more computing devices configured to perform automated operations comprising:
 calculating respective churn probability scores for each of the plurality of players; 
 identifying from the plurality of players a churn candidate by determining that the churn probability score of the churn candidate exceeds a threshold value; and 
 in response to the identifying of the chum candidate, automatically generating a player-specific message that communicates to the churn candidate one or more clues to complete an existing in-game task that is to be completed by the churn candidate by performance of gameplay actions within the first online game, and causing delivery of the message to a user device associated with the churn candidate. 
 
 
     
     
       2. The system of  claim 1 , wherein the recommendation module is configured to monitor the online engagements of the plurality of players performed during a training period of the online game. 
     
     
       3. The system of  claim 2 , wherein the recommendation module is further configured to calculate a respective online engagement score for each of the plurality of players, and to base the calculation of each churn probability score at least in part on the calculated online engagement score for the corresponding player. 
     
     
       4. The system of  claim 3 , wherein the recommendation module is configured to calculate each respective online engagement score based at least in part on a value representing how many days the corresponding player played the online game within an assessment time span. 
     
     
       5. The system of  claim 3 , wherein the recommendation module is further configured to:
 compute a group average of the calculated online engagement scores of the plurality of players; and 
 determine the threshold value for the churn probability score based at least in part on the computed group average. 
 
     
     
       6. The system of  claim 5 , wherein the recommendation module is further configured to assign a respective weight to each of the online engagement scores of the plurality of players, the group average being a weighted group average based at least in part on the respective weights of the online engagement scores. 
     
     
       7. The system of  claim 6 , wherein the recommendation module is configured such that the respective weights of the online engagement scores increase in value with an increase in recency within the training period, the group average being a weighted moving group average. 
     
     
       8. The system of  claim 7 , wherein each of the weights assigned to the respective online engagement scores of the plurality of players is an exponential weight, the group average being a weighted exponential moving group average. 
     
     
       9. The system of  claim 1 , wherein the recommendation module is further configured to identify a friend player connected to the churn candidate via a social network to help the churn candidate complete the existing task in the online game, and to automatically generate, in response to identifying the other player, a prompt message to the churn candidate for prompting the churn candidate to request help from the identified friend candidate in completing the existing in-game task. 
     
     
       10. A method comprising:
 monitoring online engagements of a plurality of players of a first online game, to determine respective engagement parameters for each player; 
 automatically calculating respective chum probability scores for each of the plurality of players; 
 automatically identifying from the plurality of players a chum candidate based at least in part on determining that the churn probability score of the churn candidate exceeds a threshold value; and 
 in an automated operation performed by one or more processors configured to perform the automated operation, in response to the identifying of the chum candidate, automatically generating a player-specific message to communicate to the churn candidate one or more clues to completing an existing in-game task that is to be completed by the chum candidate by performance of gameplay actions within the first online game, and causing delivery of the message to a user device associated with the churn candidate. 
 
     
     
       11. The method of  claim 10 , wherein the monitoring of the online engagements of the plurality of players is performed during a training period of the online game. 
     
     
       12. The method of  claim 11 , further comprising calculating a respective online engagement score for each of the plurality of players, the calculating of each churn probability score being based at least in part on the calculated online engagement score for the corresponding player. 
     
     
       13. The method of  claim 12 , wherein the calculating of each respective online engagement score is based at least in part on a value representing how many days the corresponding player played the online game within an assessment time span. 
     
     
       14. The method of  claim 12 , further comprising:
 computing a group average of the calculated online engagement scores of the plurality of players; and 
 based at least in part on the computed group average; determining the threshold value for the chum probability score. 
 
     
     
       15. The method of  claim 14 , further comprising assigning a respective weight to each of the online engagement scores of the plurality of players, the group average being a weighted group average based at least in part on the respective weights of the online engagement scores. 
     
     
       16. The method of  claim 15 , wherein the respective weights of the online engagement scores increase in value with an increase in recency within the training period, the group average being a weighted moving group average. 
     
     
       17. The method of  claim 16 , wherein each of the weights assigned to the respective online engagement scores of the plurality of players is an exponential weight, the group average being a weighted exponential moving group average. 
     
     
       18. The method of  claim 1 , further comprising identifying a friend player connected to the churn candidate via a social network to help the churn candidate complete the existing task in the online game, and, in response to identifying the friend player, automatically generating a message to the churn candidate for prompting the churn candidate to request help from the identified friend candidate in completing the existing in-game task. 
     
     
       19. A non-transitory computer readable storage medium storing instructions for causing a machine, in response to execution of the instructions by the machine, to perform operations comprising:
 monitoring online engagements of a plurality of players of a first online game, to determine respective engagement parameters for each player; 
 automatically calculating respective chum probability scores for each of the plurality of players; 
 automatically identifying from the plurality of players a churn candidate based at least in part on determining that the churn probability score of the churn candidate exceeds a threshold value; and 
 in an automated operation performed by one or more processors configured to perform the automated operation, in response to the identifying of the chum candidate, automatically generating a player-specific message to communicate to the churn candidate one or more clues to completing an existing in-game task that is to be completed by e chum candidate by performance of gameplay actions within the first online game, and causing delivery of the message to a user device associated with the churn candidate. 
 
     
     
       20. The system of  claim 1 , wherein the recommendation module is further configured to:
 identify a new in-game task for the churn candidate such that the new in-game task is easier than the existing task of the churn candidate in the online game, and 
 send to the churn candidate a prompt message with respect to the new in-game task. 
 
     
     
       21. The method of  claim 10 , further comprising:
 identifying a new in-game task for the churn candidate such that the new in-game task is easier than the existing task of the churn candidate in the online game; and 
 sending to the churn candidate a prompt message with respect to the new in-game task.